converse-mcp-server 1.15.1 → 1.17.0

package/README.md

@@ -209,8 +209,9 @@ SUMMARIZATION_MODEL=gpt-5-nano        # Default: gpt-5-nano
 - **gpt-5**: Latest flagship model (400K context, 128K output) - Superior reasoning, code generation, and analysis
 - **gpt-5-mini**: Faster, cost-efficient GPT-5 (400K context, 128K output) - Well-defined tasks, precise prompts
 - **gpt-5-nano**: Fastest, most cost-efficient GPT-5 (400K context, 128K output) - Summarization, classification
+- **gpt-5-pro**: Most advanced reasoning model (400K context, 272K output) - Hardest problems, extended compute time (EXPENSIVE)
 - **o3**: Strong reasoning (200K context)
-- **o3-mini**: Fast O3 variant (200K context)  
+- **o3-mini**: Fast O3 variant (200K context)
 - **o3-pro**: Professional-grade reasoning (200K context) - EXTREMELY EXPENSIVE
 - **o3-deep-research**: Deep research model (200K context) - 30-90 min runtime
 - **o4-mini**: Latest reasoning model (200K context)

package/docs/PROVIDERS.md

@@ -9,6 +9,7 @@ This guide documents all supported AI providers in the Converse MCP Server and t
 - **Get Key**: [platform.openai.com/api-keys](https://platform.openai.com/api-keys)
 - **Environment Variable**: `OPENAI_API_KEY`
 - **Supported Models**:
+  - `gpt-5`, `gpt-5-mini`, `gpt-5-nano`, `gpt-5-pro` - GPT-5 family with advanced reasoning
   - `o3`, `o3-mini`, `o3-pro` - Advanced reasoning models
   - `o4-mini` - Latest fast reasoning model
   - `gpt-4.1` - Large context (1M tokens)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "converse-mcp-server",
-  "version": "1.15.1",
+  "version": "1.17.0",
   "description": "Converse MCP Server - Converse with other LLMs with chat and consensus tools",
   "type": "module",
   "main": "src/index.js",
@@ -94,22 +94,22 @@
     ".env.example"
   ],
   "dependencies": {
-    "@anthropic-ai/sdk": "^0.60.0",
-    "@google/genai": "^1.15.0",
-    "@mistralai/mistralai": "^1.9.18",
-    "@modelcontextprotocol/sdk": "^1.17.4",
+    "@anthropic-ai/sdk": "^0.65.0",
+    "@google/genai": "^1.22.0",
+    "@mistralai/mistralai": "^1.10.0",
+    "@modelcontextprotocol/sdk": "^1.19.1",
     "cors": "^2.8.5",
-    "dotenv": "^17.2.1",
+    "dotenv": "^17.2.3",
     "express": "^5.1.0",
-    "lru-cache": "^11.1.0",
-    "openai": "^5.15.0",
+    "lru-cache": "^11.2.2",
+    "openai": "^6.1.0",
     "p-limit": "^4.0.0",
-    "vite": "^7.1.3"
+    "vite": "^7.1.9"
   },
   "devDependencies": {
     "@vitest/coverage-v8": "^3.2.4",
-    "cross-env": "^10.0.0",
-    "eslint": "^9.34.0",
+    "cross-env": "^10.1.0",
+    "eslint": "^9.36.0",
     "prettier": "^3.6.2",
     "rimraf": "^6.0.1",
     "vitest": "^3.2.4"

package/src/prompts/helpPrompt.js

@@ -141,12 +141,12 @@ ${formatProviderModels('OpenRouter', allModels.openrouter)}
 
 ### For Large Context Windows
 - **1M+ Tokens**: gpt-4.1 (1M), all Gemini models (1M)
-- **400K Tokens**: gpt-5 family (gpt-5, gpt-5-mini, gpt-5-nano)
+- **400K Tokens**: gpt-5 family (gpt-5, gpt-5-mini, gpt-5-nano, gpt-5-pro)
 - **256K Tokens**: grok-4 series
-- **200K Tokens**: gpt-5 series, o4-mini
+- **200K Tokens**: o3 series, o4-mini
 
 ### Special Features
-- **Web Search**: gpt-5, gpt-5-mini, gpt-5 series, o4-mini, gpt-4 series, gemini models with grounding, grok-4
+- **Web Search**: gpt-5, gpt-5-mini, gpt-5-pro, o3 series, o4-mini, gpt-4 series, gemini models with grounding, grok-4
 - **Thinking Mode**: gpt-5 series (reasoning_effort), gemini models (thinking budget)
 - **Image Support**: All models except gemini-2.0-flash-lite and grok-code-fast-1
 

package/src/providers/anthropic.js

@@ -29,6 +29,23 @@ const SUPPORTED_MODELS = {
     description: 'Claude Opus 4.1 - Highest level of intelligence and capability with extended thinking',
     aliases: ['claude-opus-4-1', 'claude-4.1-opus', 'claude-4-1-opus', 'opus-4.1', 'opus-4-1', 'claude-4-opus', 'opus-4', 'opus', 'claude-opus', 'claude-opus-4', 'opus4', 'opus4.1', 'claude-opus-4.1']
   },
+  'claude-sonnet-4-5-20250929': {
+    modelName: 'claude-sonnet-4-5-20250929',
+    friendlyName: 'Claude Sonnet 4.5',
+    contextWindow: 200000,
+    maxOutputTokens: 64000,
+    supportsStreaming: true,
+    supportsImages: true,
+    supportsTemperature: true,
+    supportsWebSearch: false,
+    supportsThinking: true,
+    minThinkingTokens: 1024,
+    maxThinkingTokens: 64000,
+    timeout: 300000,
+    supports1MContext: true, // Beta 1M context support
+    description: 'Claude Sonnet 4.5 - Latest Sonnet with enhanced intelligence and optional 1M context',
+    aliases: ['claude-4.5-sonnet', 'sonnet-4.5', 'claude-sonnet-4.5', 'sonnet4.5', 'claude-sonnet-4-5']
+  },
   'claude-sonnet-4-20250514': {
     modelName: 'claude-sonnet-4-20250514',
     friendlyName: 'Claude Sonnet 4',
@@ -42,6 +59,7 @@ const SUPPORTED_MODELS = {
     minThinkingTokens: 1024,
     maxThinkingTokens: 64000,
     timeout: 300000,
+    supports1MContext: true, // Beta 1M context support
     description: 'Claude Sonnet 4 - High intelligence and balanced performance with extended thinking',
     aliases: ['claude-4-sonnet', 'sonnet-4', 'sonnet', 'claude-sonnet', 'claude-sonnet-4', 'sonnet4']
   },
@@ -357,20 +375,23 @@ export const anthropicProvider = {
     const resolvedModel = resolveModelName(model);
     const modelConfig = SUPPORTED_MODELS[resolvedModel] || {};
 
-    // Initialize Anthropic client with default headers
-    // Use both prompt caching and extended cache duration headers for 1-hour caching
-    // Set beta headers for caching
-    const betaHeaders = ['prompt-caching-2024-07-31', 'extended-cache-ttl-2025-04-11'];
-
+    // Initialize Anthropic client
     const anthropic = new Anthropic({
       apiKey: config.apiKeys.anthropic,
-      defaultHeaders: {
-        'anthropic-beta': betaHeaders.join(',')
-      },
       // Increase timeout to 20 minutes for thinking models that may take longer
       timeout: 20 * 60 * 1000
     });
 
+    // Build beta features array for the request
+    // Use both prompt caching and extended cache duration for 1-hour caching
+    const betas = ['prompt-caching-2024-07-31', 'extended-cache-ttl-2025-04-11'];
+
+    // Add 1M context beta feature if model supports it
+    if (modelConfig.supports1MContext) {
+      betas.push('context-1m-2025-08-07');
+      debugLog(`[Anthropic] Model ${resolvedModel} supports 1M context window with beta feature`);
+    }
+
     // Convert messages to Anthropic format (system messages are always cached)
     const { systemPrompt, messages: anthropicMessages } = convertMessagesToAnthropic(messages);
 
@@ -379,6 +400,7 @@ export const anthropicProvider = {
       model: resolvedModel,
       messages: anthropicMessages,
       stream,
+      betas, // Include beta features
       ...otherOptions
     };
 
@@ -404,7 +426,7 @@ export const anthropicProvider = {
       // For other models, check against max_tokens if set
       const maxTokensLimit = requestPayload.max_tokens ||
         (resolvedModel.includes('claude-opus-4') ? 32000 :
-          resolvedModel.includes('claude-sonnet-4') ? 64000 :
+          (resolvedModel.includes('claude-sonnet-4-5') || resolvedModel.includes('claude-sonnet-4')) ? 64000 :
             modelConfig.maxOutputTokens);
 
       if (thinkingBudget > 0 && thinkingBudget < maxTokensLimit) {
@@ -449,6 +471,7 @@ export const anthropicProvider = {
         max_tokens: requestPayload.max_tokens,
         thinking: requestPayload.thinking,
         temperature: requestPayload.temperature,
+        betas: requestPayload.betas,
         message_count: requestPayload.messages?.length,
         system_length: Array.isArray(requestPayload.system) ? requestPayload.system[0]?.text?.length : requestPayload.system?.length
       }, null, 2));

package/src/providers/openai.js

@@ -52,6 +52,21 @@ const SUPPORTED_MODELS = {
     description: 'Fastest, most cost-efficient GPT-5 (400K context, 128K output) - Summarization, classification',
     aliases: ['gpt5-nano', 'gpt-5nano', 'gpt 5 nano', 'gpt-5-nano-2025-08-07']
   },
+  'gpt-5-pro': {
+    modelName: 'gpt-5-pro',
+    friendlyName: 'OpenAI (GPT-5 Pro)',
+    contextWindow: 400000,
+    maxOutputTokens: 272000,
+    supportsStreaming: false,     // GPT-5 Pro doesn't support streaming
+    supportsImages: true,
+    supportsTemperature: false,   // GPT-5 models don't support temperature
+    supportsWebSearch: true,
+    supportsResponsesAPI: true,
+    supportsDeepResearch: false,  // Not a deep research model
+    timeout: 3600000, // 60 minutes - some requests may take several minutes
+    description: 'Most advanced reasoning model (400K context, 272K output) - Hardest problems, extended compute time (EXPENSIVE)',
+    aliases: ['gpt5-pro', 'gpt-5pro', 'gpt 5 pro', 'gpt-5 pro', 'gpt-5-pro-2025-10-06']
+  },
   'o3': {
     modelName: 'o3',
     friendlyName: 'OpenAI (O3)',

package/src/utils/formatStatus.js

@@ -20,12 +20,11 @@ export async function formatJobListHumanReadable(jobsList, dependencies = {}) {
   const parts = [];
 
   // Summary line - include cancelled jobs if any
-  let summaryParts = [];
+  const summaryParts = [];
   if (jobsList.summary.active_jobs > 0) summaryParts.push(`${jobsList.summary.active_jobs} active`);
   if (jobsList.summary.completed_jobs > 0) summaryParts.push(`${jobsList.summary.completed_jobs} completed`);
   if (jobsList.summary.failed_jobs > 0) summaryParts.push(`${jobsList.summary.failed_jobs} failed`);
   if (jobsList.summary.cancelled_jobs > 0) summaryParts.push(`${jobsList.summary.cancelled_jobs} cancelled`);
-  
   const summaryStr = summaryParts.length > 0 ? summaryParts.join(', ') : '0 jobs';
   parts.push(`📊 Jobs Summary: ${summaryStr}`);
 
@@ -170,11 +169,11 @@ export async function formatHumanReadableStatus(jobStatus, options = {}, depende
   } else if (jobStatus.status === 'running' && !jobStatus.accumulated_content && jobStatus.elapsed_seconds > 5) {
     // Fallback thinking status for jobs without reasoning summaries
     const thinkingTime = Math.floor(jobStatus.elapsed_seconds);
-    debugLog(`[FormatStatus] *** FALLBACK THINKING (no reasoning_summary available)`);
-    parts.push(`Thinking: Model is processing your request (${thinkingTime}s elapsed)`);
+    debugLog('[FormatStatus] *** FALLBACK THINKING (no reasoning_summary available)');
+    parts.push('Thinking: Model is processing your request (' + thinkingTime + 's elapsed)');
   }
 
-  // Generate streaming summary for running jobs if accumulated content available  
+  // Generate streaming summary for running jobs if accumulated content available
   if (jobStatus.status === 'running' && jobStatus.accumulated_content) {
     try {
       if (dependencies.config && dependencies.providers) {